The validity of progression-free survival (PFS) 2 as a surrogate endpoint for overall survival (OS) in randomized controlled trials (RCTs) of advanced solid tumors.
1516 Background: OS is the gold-standard endpoint for treatment efficacy in oncology RCTs. However, prolonged follow-up is required to obtain mature data, which impedes regulatory approval of potentially beneficial therapies. Furthermore, increasing therapeutic options including cross over to investigational agents at disease progression often confound OS findings. PFS-2, defined as time from randomization to progression on second-line therapy, has been proposed as a potential surrogate endpoint for OS. Using a meta-analytic approach, we aimed to assess the association between OS and PFS-2, and its validity compared with other surrogate endpoints. Methods: We performed an electronic literature search to identify RCTs of systemic therapies that reported PFS-2 as a pre-specified endpoint with defined follow up protocols. Articles were screened for eligibility and outcome data were extracted. Correlations in the relative treatment difference between treatment arms for OS vs PFS-2, PFS-1, and objective response rate (ORR) were assessed as the comparison of hazard ratios (HR) and odds ratio (OR) respectively. Results: 38 eligible RCTs comprising of 44 analysis units, with 19,031 patients across 8 unique tumor types were identified. The majority received targeted therapies (72.2%), followed by immunotherapy (IO; 26.3%). The correlations of HR-OS/HR-PFS-2, HR-OS/HR-PFS-1 and HR-OS/OR-ORR were r = 0.67 (95% CI 0.08-0.69), r = 0.12 (95% CI 0.00-0.13) and r = 0.21 (95% CI 0.00-0.33) respectively. Correlations between OS with surrogates was assessed in different subgroups according to post-progression survival times (SPP), types of treatment agents and rate of subsequent therapy after progression (table). Conclusions: Across diverse tumors and therapies, treatment effect on PFS-2 has modest to strong correlation with OS, but poor correlations were observed between OS-PFS-1 and OS-ORR respectively. Across all subgroups, PFS-2 performs consistently better than traditional surrogates of PFS-1 and ORR, validating and providing support for use in future RCTs.[Table: see text]